Bankers are learning some hard lessons about yet another of the crazes that have pervaded marketing and technology in recent years.

Data warehouses-giant information repositories for sales and decision support-are under development at virtually all the largest U.S. banks. The banks entered into these projects expecting such strategic and tactical benefits as the ability to analyze profitability by product and account. These analyses enable banks to design highly customized marketing programs that can segment target audiences down to a single consumer.

But the process is proving more drawn out and complex than expected. Eager to see some return on the multimillion-dollar investments, many institutions are scaling back expectations and turning their attention to smaller, more focused "data marts."

Usually designed for specific functions such as credit card marketing, data marts are meant to ease the transition to data warehouses, not replace them. Still, despite a good deal of hype by system vendors and buyers alike, industry experts contend most banks are unlikely to realize the "mega-warehouse" ideal until at least the end of the decade.

Building an enterprise-wide data warehouse "is like ending world hunger- it's a very altruistic objective, but not terribly realistic," said Aaron Zornes, executive vice president of application delivery strategies with Meta Group, a Stamford, Conn., technology research firm.

Bringing together "all the information about all your subjects so that everyone can access it from every possible view" is a good objective, he added, but probably unattainable in the short term.

The data-warehouse opportunity posed a chicken-or-egg problem: whether to build a warehouse and design applications later, or to start with desired applications and build a warehouse around those needs.

Many banks started with the former approach but switched to the latter as they turned more conservative.

"People are saying they're not going to be the first out of the gate- even the same people who two years ago were saying they were eager" to build enterprise-wide warehouses, said Mary Knox, project manager with Mentis Corp., a Raleigh, N.C., bank technology research company.

Even with the shifting winds, data warehousing remains one of the more popular areas of bank technology investment.

Mentis said banks spent $1.05 billion last year on data warehousing and related information management projects. That number is expected to rise to $1.2 billion for 1997.

But few data warehouses are yielding the kind of information reporting that many expect from them: ascertaining the likes and dislikes of individual customers, determining their profitability, and tailoring products and sales pitches to maximize that profitability.

Even though banks have all the necessary information, they "have not been very successful in mining that information to effectively understand their markets or their customers," said John Beale, executive vice president of Cast Management Consulting and City National Corp.'s interim chief information officer.

Given the ample spending in the field, there is no lack of desire or financial commitment.

Observers said it was important for bankers to take a long-term view and avoid getting discouraged by failures.

Most of the major retail banks have experienced setbacks in data warehouse construction. Several, including Wells Fargo & Co., have had to either rethink or cancel parts of their projects.

"There have been some blowups, but basically banks write them off as a learning experience," said Mr. Zornes of Meta Group. "Some people started overly ambitious, and some people started with the wrong technology."

As word spread about the mixed results, some banks may have gotten gun shy. However, the technology needed to support the warehouses is available and improving. And as vendors of both hardware and software scramble to outdo each other, banks have ever-improving leverage in negotiating on price.

Therefore, banks that have been watching and waiting may find that now is the time to dive in.

Consultants and bankers advise setting short-term expectations low. Most data warehousing projects take years to bear fruit. City National, for instance, has its project on a five-year plan.

In addition, bankers should understand that data warehousing projects never really end. Even after a basic structure is completed, the systems need constant attention and tweaking to yield useful information.

Gerald R. Modes, president and chief executive officer of Hyperparallel, a San Francisco-based data mining software company, gives this advice: "My uncle used to tell me you can't bite a watermelon. Cutting these things up into bite-size pieces is, in fact, what you have to do."

Mr. Zornes added, "It's a trite saying, but it works in data warehouse: Think global; act local."

To put it another way, he said, "Think architecture, think enterprise in the long haul-but get quick results with building something called a data mart."

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.